INDEX

Explanations

superlatives and positive descriptors

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ople

-0.81

clips

-0.76

pta

-0.72

EA

-0.69

ilus

-0.69

phrine

-0.68

mitt

-0.68

lation

-0.65

qqa

-0.65

earance

-0.65

POSITIVE LOGITS

sword

1.07

 strides

0.90

 apes

0.81

axe

0.81

 enthus

0.80

 Dane

0.80

 opportunity

0.78

 outdoors

0.77

 distances

0.77

itud

0.77

Activations Density 0.377%

superlatives and positive descriptors

No Comments

No Known Activations